When the name “Black Swan” appears in a book title, one must wonder if the publisher is merely attempting to latch onto the coattails of Nassim Nicholas Taleb’s very successful 2007 study of the cataclysmic impact of improbable events, The Black Swan: The Impact of the Highly Improbable.
However, Kenneth Posner, author of Stalking The Black Swan and a veteran Wall Street analyst, is a true adherent. His first footnote cites the famous Jorion-Taleb debate of 1997, in which Taleb attacked the popular “Value At Risk” methodology, not only as impractical, but also as dangerous.
“My refutation of VAR,” wrote Taleb, “does not mean that I am against quantitative risk management—having spent most of my adult life as a quantitative trader, I learned the hard way the pitfalls of such methods. I am simply against the application of unseasoned quantitative methods.” Philippe Jorion, professor of finance at the University of California, conceded many of Taleb’s points in his reply. (You can read the debate at www.DerivativesStrategy.com.)
It is in that spirit that Posner continues the attack on the “Emperor’s New Clothes,” quantitative investment techniques.
Author’s plan of attack
Posner organizes his thoughts about “Research and Decision Making in a World of Extreme Volatility” (the book’s subtitle) into three areas: uncertainty, information, and judgment. His thesis is simple: We will never be able to fully understand our world, so we should overlay judgment when using today’s dazzling computerized trading and investment tools to penetrate the avalanche of data in financial markets.
Rules derived from complex models, Posner argues, cannot substitute for sound analysis. In fact, blind reliance on models can make us complacent and thus more vulnerable to devastating Black Swan surprises.
This is a straightforward thesis—and quite refreshing in a field too-often characterized by algorithmic how-to books for technically-oriented investment analysts.
Painful lessons from the Crisis
Posner balances recent research into the limitations of our ability to cope with information overload and rapid-fire market movements with practical examples of “phase shifts” in the beliefs among investor groups. Throughout the book, Posner uses himself and his research team as his primary subjects. Posner examines (and exposes) his own analytic mistakes in a series of case studies which follow one of Taleb’s main trading rules: “Read every book by traders to study where they lost money. You will learn nothing relevant from their profits (the markets adjust). You will learn from their losses.”
Posner has drawn cases from the research papers he wrote over 15 years as a securities analyst at Morgan Stanley. As head of “specialty finance,” Posner led a research team which covered housing finance, including Fannie Mae, Freddie Mac, and Countrywide; student lending at Sallie Mae; as well as consumer finance at MasterCard, American Express, and Discover. When describing market volatility, Posner speaks with the savvy of one who carried coffee from the subway to his desk as the biggest names in the financial sector imploded almost every week from the summer of 2007 through late 2008.
Overwhelming force of feedback loops and catalysts
Posner points out that quantitative trading strategies were widely copied after 2000, to the point that in August 2007, when many believe the financial crisis really began, a feedback loop had arisen to magnify the broader market effect of losses in the subprime lending sector.
Citing research by academics at MIT, Posner explains that the investment banks and hedge funds in the subprime sector tried to survive by withdrawing positions from other sectors of the market, creating losses for those investors who remained. Like a fishing boat in a rogue wave, once-profitable strategies were suddenly overwhelmed by a contagion of selling in all trading sectors.
Posner believes that collective action of that nature creates non-linear forces that make successful forecasting nearly impossible. Traders who normally follow strategies that tend to offset each other unknowingly act in unison as their views of the crisis converge. Leverage, he points out, intensifies the feedback effect and adds stress to investors and traders alike.
Posner applies his views to individual securities as well as for market sectors. He recommends that fundamental investors, meaning those who analyze the factors affecting a company’s earnings, continually search for catalysts that will trigger a sudden convergence of opinion among other investors. As an illustration, Posner cites as catalyst the loss of confidence by foreign central banks in Fannie Mae’s debt that drove the mortgage insurer’s stock price, in under three months, to zero from $20. The result was a “black swan” pricing event after a wave of selling by shocked investors.
Modeler’s Achilles heel: correlation vs. causality
Posner also recounts the travails of Providian, which fell 96% during four months of 2001.
He and his team had weighted the three variables most influential, they believed, in affecting the earnings of the credit card issuer: the economy, the credit cycle, and the Providian business model. But, in retrospect, Posner came to realize that the economy and the credit cycle were too closely correlated in Providian’s business model, so the systemic effect of the post-dotcom downturn on the stock was much greater than expected. He also realized that the true causative variable was Providian’s inability to respond to competitive pressures once its own aggressive marketing programs were curtailed by regulators. His team correctly predicted the slowdown in Providian’s results, but not its ultimate collapse. For that insight, he says he should have focused on the essential causality in competitive factors.
Posner devotes fully one-third of his book to an outline of techniques intended to help us recognize the limits of today’s analytic tools and avoid the overconfidence that results from excessive reliance on complex models. In one instance, he praises the value of Monte Carlo simulations, but then cautions that these recursive estimations “can be extremely dangerous with the wrong correlation assumptions.”
This is a valuable insight, especially when introduced within the context of 2007’s MasterCard public offering, where his team used a Monte Carlo simulation to combine six seemingly uncorrelated variables to estimate future earnings and resulting stock prices. While their forecasts were relatively accurate, it turned out that management incentives were the true causative variable. In another example, Posner shows how political risk was the real uncorrelated driver of Fannie Mae’s stock price when foreign central banks refused to roll over the mortgage insurer’s debt.
How cognitive dissonance sets up the shock
Forecasting the subprime collapse was nearly impossible, writes Posner, given the inadequacy of the data and models available to analysts in 2006. He now thinks “the properly specified model for subprime mortgage losses should have included variables for global liquidity, competitive pressure on lenders, and the risk of discontinuous shocks to home prices—and the circular logic between home prices and credit losses. Yet readily available data did not exist for these factors. Nor is there straightforward math to link them.”
At that time, Posner asserts, analysts would have had to run better models more frequently and with ever-newer data to get even close to a forecast for subprime losses. Yet, even having the right models would not have been a complete solution to the risk management problem. Conditions were changing in a way that induces resistance to re-thinking by observers. Posner cites a condition called “cognitive dissonance” by experts, which arises when humans persist in a first, but largely inaccurate mindset even when conditions are undermining their original assumptions.
Using interviews to get past scripts
Posner enriches his parables with analytic tricks of the trade. For example, he explains how a good analyst can interview a CEO in such a way as to get behind the formulaic response of the investor relations department.
“The art of interviewing,” he writes, “is to maneuver the manager into a spot where she cannot credibly deliver a scripted answer and where even a non-answer would have diagnostic power for critical issues.”
Posner recommends that analysts: 1. control the agenda; 2. ask specific questions; 3. aim questions at the “how” and “why”; 4. recognize non-answers; and 5. be wary of revealing their own views.
As always, Posner gives compelling examples to support his views. These rules are just as helpful for CEOs in recognizing the pattern of questions before falling into the analyst’s gotcha question.
Posner’s examples are intended to foster caution in the use of models—a good lesson as both financial regulators and their regulated entities are moving inevitably toward even more complexity in their formulation and application. Ultimately, Posner concludes that judgment and expert advice must be used to refine the computer’s analytic output, especially in the kind of volatile markets that have appeared so frequently of late. As an extra benefit, his narrative examples will refine the lessons learned by those who were at the center of the crisis’ storms, while revealing to less-familiar readers the sincere though often-futile diligence of those who tried to steer through the hazards of the financial tsunami of the last few years.
If you'd like to review books for our online book column, or have recently read a book that you found helpful that we haven't already reviewed, please e-mail firstname.lastname@example.org
Tagged under Books for Bankers,